Author Topic: GUI segmentation fault on startup  (Read 13590 times)

smokeandmirrors

  • Jr. Member
  • **
  • Posts: 2
GUI segmentation fault on startup
« on: June 02, 2008, 04:57:53 PM »
Hi guys, we've just received a new RHEL WS4 Update5 workstation and have installed qube, only to find the GUI tool crashes on startup:

Code: [Select]
# /usr/local/pfx/qube/bin/qube
      4459: p???Segmentation fault

And a full backtrace looks like:

Code: [Select]
GNU gdb Red Hat Linux (6.3.0.0-1.143.el4rh)
Copyright 2004 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "x86_64-redhat-linux-gnu"...Using host libthread_db library "/lib64/tls/libthread_db.so.1".

(gdb) bt
No stack.
(gdb) run
Starting program: /usr/local/pfx/qube/bin/qube
Reading symbols from shared object read from target memory...done.
Loaded system supplied DSO at 0xffffe000
[Thread debugging using libthread_db enabled]
[New Thread 4160526016 (LWP 4565)]
[New process 4565]
      4565: `???
Program received signal SIGSEGV, Segmentation fault.
0x0031a2bb in ?? ()
(gdb) bt
#0  0x0031a2bb in ?? ()
#1  0x003addc1 in ?? ()
#2  0x00000001 in ?? ()
#3  0xffffbf20 in ?? ()
#4  0xffffbf44 in ?? ()
#5  0x008e9ab9 in do_lookup_x () from /lib/ld-linux.so.2
#6  0x00a29cb8 in dlopen_doit () from /lib/libdl.so.2
#7  0x008ed4ee in _dl_catch_error () from /lib/ld-linux.so.2
#8  0x00a2a2bb in _dlerror_run () from /lib/libdl.so.2
#9  0x00a29d11 in dlopen@@GLIBC_2.1 () from /lib/libdl.so.2
#10 0x080be3b2 in _PyImport_GetDynLoadFunc ()
#11 0x080b510f in _PyImport_LoadDynamicModule ()
#12 0x080b2ee5 in load_module ()
#13 0x080b3343 in import_submodule ()
#14 0x080b3575 in load_next ()
#15 0x080b39da in PyImport_ImportModuleEx ()
#16 0x080df18e in builtin___import__ ()
#17 0x080de0e1 in PyCFunction_Call ()
#18 0x08059727 in PyObject_Call ()
#19 0x0809a590 in PyEval_CallObjectWithKeywords ()
#20 0x0809bf97 in eval_frame ()
#21 0x080a0486 in PyEval_EvalCodeEx ()
#22 0x080a074d in PyEval_EvalCode ()
#23 0x080b181b in PyImport_ExecCodeModuleEx ()
#24 0x080b1939 in PyImport_ExecCodeModule ()
#25 0x080566b4 in importModules () at ../common/launch.c:609
#26 0x08056bf3 in doIt (argc=1, argv=0xffffda54) at ../common/launch.c:1018
#27 0x080557d0 in main (argc=1, argv=0xffffda54) at main.c:93

Do you have any ideas as to what might be missing from the workstation's environment to get the GUI working?

Thanks,
Nick

« Last Edit: June 02, 2008, 04:59:59 PM by smokeandmirrors »

Scot Brew

  • Hero Member
  • *****
  • Posts: 272
    • PipelineFX
Re: GUI segmentation fault on startup
« Reply #1 on: June 03, 2008, 12:46:48 AM »
Yes, it appears that there may be a library incompatibility with your setup, though we have many customers using RHEL4 with Qube and the QubeGUI.  Here are a few things you can try to track down what you are seeing in your environment:
  • Launch python from the commandline and import the qb python module:
Code: [Select]
python
import qb
    If there is something preventing the loading of the qb python module, then it should error here.  If successful, then this eliminates the qb python module having loading issues.
    [/li]
  • Verify that you have installed the RHEL4 rpms (for 32-bit or 64-bit depending on your system).  We have run across this one before that the wrong rpm was installed. It usually results in a segmentation fault.  You may want to uninstall on that machine and reinstall just to be sure the right rpms were added.

smokeandmirrors

  • Jr. Member
  • **
  • Posts: 2
Re: GUI segmentation fault on startup
« Reply #2 on: June 03, 2008, 03:20:09 PM »
Is the qb module meant to be in the default python import path? I can see one in the api/ subdirectory in qube, but nothing that's installed system-wide.

In any case, the binary .so loads without a problem, which leads us to think that it's a package issue with RPMs.  We're using this as a 64bit install, so I'm erring towards the conclusion that the 32bit environment is incomplete in some way. RHEL4 makes that hard since the architecture isn't presented in the RPM package name, as it is in the later Fedora releases.


Scot Brew

  • Hero Member
  • *****
  • Posts: 272
    • PipelineFX
Re: GUI segmentation fault on startup
« Reply #3 on: June 16, 2008, 10:59:29 PM »
We have been able to reproduce this QubeGUI segmentation fault issue on RHEL4-x64 and are working on getting out a compiled QubeGUI rpm specific to that platform.

We have verified that the QubeGUI can be run interactively using the standard python 2.3 that comes with the OS, the qb python module for RHEL4-x64, and the latest wxPython for RHEL4-x64. This can be done while we work on getting out a compiled QubeGUI rpm.

Steps to install/run QubeGUI interactively on RHEL4-x64
  • Install yum  (makes it easier to install wxPython and its dependencies)
  • Install wxPython (the rpms have been found at the Pramberger repository)
  • Install numpy
  • Add qb python module to the PYTHONPATH
    setenv PYTHONPATH /usr/local/pfx/qube/api/python/
  • Run the QubeGUI
    /usr/bin/python /usr/local/pfx/qube/api/python/qb/gui/qube.py

Scot Brew

  • Hero Member
  • *****
  • Posts: 272
    • PipelineFX
Re: QubeGUI 5.3 compiled for RHEL4-x64
« Reply #4 on: June 23, 2008, 08:37:49 PM »
A compiled version of the QubeGUI 5.3 for Red Hat Enterprise Linux 4 64-bit (RHEL4-x64) has been made available.  It can be downloaded from our ftp server at:
   
    ftp://ftp.pipelinefx.com/pub/qubegui/5.3/current

Installing this compiled version means that one does not have to install wxPython or any other dependencies to launch the GUI.